Passive enrollment method for speaker identification systems

Abstract

Techniques for passive enrollment of a user in a speaker identification (ID) device are provided. One technique includes: parsing, by a processor of the speaker ID device, a speech sample, spoken by the user, into a keyword phrase sample and a command phrase sample; identifying, by a text-dependent speaker ID circuit of the speaker ID device, the user as the speaker of the keyword phrase sample; associating the command phrase sample with the identified user; determining if the command phrase sample in conjunction with one or more earlier command phrase samples associated with the user is sufficient command phrase sampling to enroll the user in a text-independent speaker ID circuit of the speaker ID device; and enrolling the user in the text-independent speaker ID circuit using the command phrase samples associated with the user after determining there is sufficient command phrase sampling to enroll the user.

What is claimed is: 
     
       1. A method for enrollment of a user in a speaker identification (ID) device, the method carried out by the speaker ID device and comprising:
 actively enrolling the user using one or more samples of a keyword phrase spoken by the user; 
 parsing a user speech sample, spoken by the user, into a command phrase sample and a keyword phrase sample; 
 identifying the user as the speaker of the keyword phrase sample; 
 associating the command phrase sample with the identified user; and 
 passively enrolling the user using the command phrase sample in conjunction with one or more earlier command phrase samples associated with the user, the command phrase sample and the one or more earlier command phrase samples together making up at least two command phrase samples associated with the user. 
 
     
     
       2. The method of  claim 1 , wherein prior to passively enrolling the user, the method comprises:
 saving the at least two command phrase samples associated with the user to a storage; and 
 in response to determining the stored command phrase samples associated with the user are sufficient command phrase sampling, allowing the passive enrollment to commence. 
 
     
     
       3. The method of  claim 2 , wherein the sufficient command phrase sampling comprises at least one minute of speech. 
     
     
       4. The method of  claim 2 , wherein in response to determining the stored command phrase samples associated with the user are insufficient command phrase sampling, postponing the passive enrollment. 
     
     
       5. The method of  claim 1 , wherein the user speech sample is uttered as a command to the speaker ID device from the user, the method further comprising sensing the user speech sample being uttered as the command to the speaker ID device from the user. 
     
     
       6. The method of  claim 1 , wherein the user is a first user, the method further comprising:
 actively enrolling a second user using one or more samples of the keyword phrase spoken by the second user; 
 parsing a user speech sample, spoken by the second user, into a second command phrase sample and a second keyword phrase sample; 
 identifying the second user as the speaker of the second keyword phrase sample; 
 associating the second command phrase sample with the identified second user; and 
 passively enrolling the second user using the second command phrase sample in conjunction with one or more earlier command phrase samples associated with the second user, the second command phrase sample and the one or more earlier command phrase samples associated with the second user together making up at least two command phrase samples associated with the second user. 
 
     
     
       7. The method of  claim 6 , further comprising:
 saving the at least two command phrase samples associated with the first user and the at least two command phrase samples associated with the second user to a storage; 
 in response to determining the stored command phrase samples associated with the first user are sufficient command phrase sampling of the first user, allowing the passive enrollment of the first user to commence; 
 in response to determining the stored command phrase samples associated with the first user are insufficient command phrase sampling of the first user, postponing the passive enrollment of the first user; 
 in response to determining the stored command phrase samples associated with the second user are sufficient command phrase sampling of the second user, allowing the passive enrollment of the second user to commence; and 
 in response to determining the stored command phrase samples associated with the second user are insufficient command phrase sampling of the second user, postponing the passive enrollment of the second user, 
 wherein the sufficient command phrase sampling of the first user comprises at least 20 command phrase samples, and the sufficient command phrase sampling of the second user comprises at least 20 command phrase samples. 
 
     
     
       8. A speaker identification (ID) device, the device comprising:
 a text-independent speaker ID circuit configured to receive a compound speech sample spoken by a user; 
 a speech parsing circuit configured to parse the received compound speech sample into a command phrase sample and a keyword phrase sample; 
 a text-dependent speaker ID circuit configured to identify the user as the speaker of the keyword phrase sample; 
 a text-dependent enrollment circuit configured to enroll the user in the text-dependent speaker ID circuit using one or more samples of the keyword phrase spoken by the user; 
 a command phrase association circuit configured to associate the command phrase sample with the identified user; and 
 a command phrase enrollment testing circuit configured to determine if the command phrase sample in conjunction with one or more earlier command phrase samples associated with the user is sufficient command phrase sampling to enroll the user in the text-independent speaker ID circuit, the command phrase sample and the one or more earlier command phrase samples together making up at least two command phrase samples associated with the user. 
 
     
     
       9. The speaker ID device of  claim 8 , further comprising a text-independent enrollment circuit configured to enroll the user in the text-independent speaker ID circuit using the at least two command phrase samples associated with the user after the command phrase enrollment testing circuit determines there is sufficient command phrase sampling to enroll the user in the text-independent speaker ID circuit, wherein the text-independent speaker ID circuit is further configured to identify the user as the speaker of a second speech sample spoken by the user after the user is enrolled in the text-independent speaker ID circuit. 
     
     
       10. The speaker ID device of  claim 8 , wherein the text-dependent enrollment circuit is further configured to enroll one or more other users in the text-dependent speaker ID circuit using a corresponding one or more samples of the keyword phrase spoken by each of the other users. 
     
     
       11. The speaker ID device of  claim 8 , wherein the compound speech sample is uttered as a command to the speaker ID device from the user, the speaker ID device further comprising a microphone configured to sense the compound speech sample being uttered as the command to the speaker ID device from the user. 
     
     
       12. A speaker ID system comprising a plurality of speaker ID devices including the speaker ID device of  claim 8 , wherein after the user is passively enrolled, the system is configured to: receive a query from the user; determine which speaker ID device of the plurality is closest to the user at a given time the query is received; and respond to the query via the speaker ID device of the plurality that is closest to the user. 
     
     
       13. A computer program product including one or more non-transitory machine-readable mediums encoded with instructions that when executed by one or more processors cause a process to be carried out for enrollment of a user in a speaker identification (ID) device, the process comprising:
 actively enrolling the user using one or more samples of a keyword phrase spoken by the user; 
 parsing a user speech sample, spoken by the user, into a command phrase sample and a keyword phrase sample; 
 identifying the user as the speaker of the keyword phrase sample; 
 associating the command phrase sample with the identified user; and 
 passively enrolling the user using the command phrase sample in conjunction with one or more earlier command phrase samples associated with the user, the command phrase sample and the one or more earlier command phrase samples together making up at least two command phrase samples associated with the user. 
 
     
     
       14. The computer program product of  claim 13 , wherein prior to passively enrolling the user, the process comprises:
 saving the at least two command phrase samples associated with the user to a storage; and 
 in response to determining the stored command phrase samples associated with the user are sufficient command phrase sampling, allowing the passive enrollment to commence. 
 
     
     
       15. The computer program product of  claim 14 , wherein the sufficient command phrase sampling comprises at least one minute of speech. 
     
     
       16. The computer program product of  claim 14 , wherein in response to determining the stored command phrase samples associated with the user are insufficient command phrase sampling, postponing the passive enrollment. 
     
     
       17. The computer program product of  claim 13 , wherein the user is a first user, the process further comprising:
 actively enrolling a second user using one or more samples of the keyword phrase spoken by the second user; 
 parsing a user speech sample, spoken by the second user, into a second command phrase sample and a second keyword phrase sample; 
 identifying the second user as the speaker of the second keyword phrase sample; 
 associating the second command phrase sample with the identified second user; and 
 passively enrolling the second user using the second command phrase sample in conjunction with one or more earlier command phrase samples associated with the second user, the second command phrase sample and the one or more earlier command phrase samples associated with the second user together making up at least two command phrase samples associated with the second user. 
 
     
     
       18. The computer program product of  claim 17 , the process further comprising:
 saving the at least two command phrase samples associated with the first user and the at least two command phrase samples associated with the second user to a storage; 
 in response to determining the stored command phrase samples associated with the first user are sufficient command phrase sampling of the first user, allowing the passive enrollment of the first user to commence; 
 in response to determining the stored command phrase samples associated with the first user are insufficient command phrase sampling of the first user, postponing the passive enrollment of the first user; 
 in response to determining the stored command phrase samples associated with the second user are sufficient command phrase sampling of the second user, allowing the passive enrollment of the second user to commence; and 
 in response to determining the stored command phrase samples associated with the second user are insufficient command phrase sampling of the second user, postponing the passive enrollment of the second user, 
 wherein the sufficient command phrase sampling of the first user comprises at least 20 command phrase samples, and the sufficient command phrase sampling of the second user comprises at least 20 command phrase samples. 
 
     
     
       19. A system comprising the computer program product of  claim 13 , wherein after the user is passively enrolled, the system is configured to: receive a query from the user; determine which speaker ID device of a plurality of speaker ID devices is closest to the user at a given time the query is received; and respond to the query via the speaker ID device of the plurality that is closest to the user.